The Volkswagen Touareg fielded by California's Stanford University on Sunday bagged the purse awarded by the US Defence Department to help enhance development of battlefield robots, both Stanford and the Pentagon said.
The car, named Stanley, was one of five unmanned and self-controlled vehicles to cross the finish line in the 210.5km Defence Advanced Research Projects Agency (DARPA) Grand Challenge Course, held in the Mojave Desert in the US state of Nevada at the weekend.
"These vehicles haven't just achieved world records, they've made history," said DARPA director Tony Tether after Stanley bagged the big prize by completing the gruelling course in six hours, 53 minutes and 58 seconds.
The modified diesel-powered Volkswagen Touareg R5, controlled by a space-age drive-by-wire system using six powerful Pentium M computers, travelled at an average speed of 30.5km an hour.
"We have done the impossible. People said, 'give up its not possible', but we did it," said Stanford University artificial intelligence expert Sebastian Thrun, one of the computer wizards who developed Stanley over 14 months.
Experts say driverless cars will
win back time lost stuck in traffic
"I honestly believe that the impact of this and other things related to self-driving cars could turn out to be more fundamental to mankind than the invention of the internet," German-born Thrun told AFP.
In addition to allowing the military to send machines into battle while keeping soldiers safe, driverless vehicles will free up millions of hours of civilian manpower annually as commuters win back time lost in traffic, he said.
"That time is just being wasted. When cars can drive themselves, people will be considerably safer and freed of the burden of driving, making them productive while they are in the car because they can do other things," he said.
The development will allow people to use robotic cars like trains as they would be able to drop one person off, before returning to pick up other commuters.
"The problem with cars now is that they spend the vast majority of their time parked in the wrong location so they cannot be used by other drivers," Thrun said.
'Stanley' averaged a speed of
30.5km an hour
And the technological advance could save up to 43,000 lives lost on US roads in accidents each year.
"Self-driving cars will be much safer as they will not be subject to human error as they are now," the computer expert said.
Stanley, developed by Stanford's Artificial Intelligence Laboratory and Volkswagen, was one of 23 cars competing in the second Grand Challenge across rugged desert and the first of only five to cross the finish line.
In the first robotic car competition last year, none of the entrants finished the course to claim the $1 million price. In fact, none went farther than 12.8km.
Also finishing was Carnegie Mellon's Red Team Toos Highlander, another military-designed Humvee, which finished close behind with a time of 7:14:00, averaging 29.1km an hour.
The Gray Teams KAT-5 finished at 7:30:16, while Oshkosh Trucks 16-tonne robot, TerraMax, also finished the course with an average speed of 28km an hour.
The truck, however, was not eligible for the prize offered by the Pentagon to spur robotic development as its finishing time exceeded the 10-hour limit by about two hours.
"Self-driving cars will be much safer as they will not be subject to human error as they are now"
Most of the competing vehicles guided themselves over the winding and obstacle-strewn terrain using laser sensors, a Global Positioning System, radar and cameras.
Thrun praised the US military for funding the project, but said that the ramifications of driverless cars would reach far beyond the military - and relatively soon.
"I believe that within 50 years or maybe even 30 we will see this new technology in our everyday lives," the computer scientist said.